Infrastructure archaeology and other constraints
I'm Ryan Nelson. My career took off in the mid-90s building systems that weren't expected to scale—but did. If you're from NYC, I may have been your dialup sysadmin way back in the day.
I spent a decade helping build MLB.com for Major League Baseball, then moved west to build clouds in California.
After semi-retiring to Seattle and taking a pandemic pause, some old industry peers convinced me to come run
Tech at TED. I love my job. I used to be root for the home team ⚾; now I'm part of the "T" in TED.
-
Intrusion Countermeasures, Recreationally
February 11, 2026There's a line people like to repeat about cyberpunk. William Gibson imagined cyberspace, simstim, decks, the Sprawl. He missed the cell phone, but he got the mood exactly right. The network is alive. It's hostile in a curious way. What's been interesting about running local AI lately is that it feels like play again.
Read more → -
My AI Agent Has a Kill Switch Now
January 31, 2026Like many dotcom Gen-X hipsters this week, I'm playing with OpenClaw - an AI agent in a Slack channel where I'm the only member. I can reach it from my phone. It has shell access to a Linux workstation in my barn. Also terrifying. So I built otp-challenger - TOTP verification for AI agents.
Read more → -
Stealing Fire: Why My New Open Source Tools Borrow from Hackers
January 20, 2026I came to New York City in 1995. The dotcom explosion was happening. I was working as a sysadmin at an ISP. Here is a statistic that might terrify a modern Hiring Manager: I have never held a job where I did not have "root." This is "gonzo-sysops." But you learn things when you are five hops deep in a server that is currently on fire.
Read more → -
The Desire Path: Teaching 2025 AI to Survive 1990s Infrastructure
January 18, 2026There is a specific kind of intellectual satisfaction in building a ship in a bottle. For the last few months, this has been my reality with AI—asking Claude to SSH into jump hosts, launch tmux sessions, and type commands into shells. It turns out that modern Large Language Models have zero intuition for "terminal physics".
Read more →